Only the 1st couple sentences are being captured into the dataset for community descriptions. I suppose this is because the Lemmyverse UI only shows the first couple sentences, which I suspect is due to limited screen space.

It would be better for searching purposes to capture the full descriptions in the dataset. From there, once the dataset has more complete descriptions, they could also be used to produce better summaries than just taking the 1st couple sentences. Consider this tool:

https://goblin.tools/Formalizer

You feed it text, the AI bot absorbs it all, and you can order a short version that captures the important points. The short version could then also be printed in a smaller font to conserve screen real estate.

It’s probably not great to rely on a 3rd party service like goblin.tools. Ideally there would be some FOSS that does this job.

Ideally in the end, the full original description is used for searches and then the short summary is used for displaying results.